New approaches to providing nutritional information 173 7. 4.5 Relevant indices that are based on factors that confer relevance on food data The relevance of food information is determined by validity, sufficiency, practi cality, and communicability. Is an index a true reflection of a change in a bio- marker or end-point, is it sufficient on its own to predict a change in the end-point, is it a variable that can be measured easily, and expressed in terms that users understand well enough to use in food choice? 7.4.6 Food data that is easily understood Food data is not relevant if it cannot accurately link consumer behaviour to health end-points, in other words, if it cannot guide food choice for health. To do so it should be easily used. The relative efficacy of foods may, for instance, be expressed in terms of equivalents to a familiar reference that exhibits a specifie effect to a known degree, as in wheat bran equivalents and faecal bulking. Gly chemic index(GD), on the other hand, is an example of a number that is supposed to represent the glycaemic potency of a food. However, unlike intake of a nutri- ent, GI does not change with the composition, serving size, or intake of food, so it makes little sense to consumers, and cannot be used accurately to modify eating patterns that affect blood glucose. 4 7.5 Limitations of food composition data: the case of carbohydrates The above framework for building practical, evidence-based data sets linked to health end-points is illustrated below by reference to two physiological effects of food carbohydrates: postprandial glycaemia(post-meal elevation of blood glucose), and faecal bulking. Postprandial glycaemia is determined largely by carbohydrate digestibility, and faecal bulk largely by non-digestible, non- fermentable polysaccharides. 7.5.1 Limitations of carbohydrate composition data Standard food analyses do not account for the large effects of the structure of car- bohydrate molecules and foods in the carbohydrate nutrition. Monosaccharide composition and order, glycosidic bonds, degree of polymerisation, chain con figurations, non-covalent interactions between chains, and crosslinks that carbo- hydrates readily form may all greatly affect physicochemical properties, 4and the physiological effects that depend on such properties. Furthermore, food struc ture, such as particle size, may considerably modulate the ability of food carbo- hydrates to express their potential properties, by limiting solubility, extraction, and access of dis enzymes. Beyond effects on extraction, interactions between carbohydrates and other food components in the intestine are multiple nd complex. The amounts of carbohydrate fractions in foods are therefore not usually reliable guides to their physiological effectiveness7.4.5 Relevant indices that are based on factors that confer relevance on food data The relevance of food information is determined by validity, sufficiency, practi￾cality, and communicability. Is an index a true reflection of a change in a bio￾marker or end-point, is it sufficient on its own to predict a change in the end-point, is it a variable that can be measured easily, and expressed in terms that users understand well enough to use in food choice? 7.4.6 Food data that is easily understood Food data is not relevant if it cannot accurately link consumer behaviour to health end-points, in other words, if it cannot guide food choice for health. To do so it should be easily used. The relative efficacy of foods may, for instance, be expressed in terms of equivalents to a familiar reference that exhibits a specified effect to a known degree, as in wheat bran equivalents and faecal bulking.42 Gly￾caemic index (GI), on the other hand, is an example of a number that is supposed to represent the glycaemic potency of a food.43 However, unlike intake of a nutri￾ent, GI does not change with the composition, serving size, or intake of food, so it makes little sense to consumers, and cannot be used accurately to modify eating patterns that affect blood glucose.44 7.5 Limitations of food composition data: the case of carbohydrates The above framework for building practical, evidence-based data sets linked to health end-points is illustrated below by reference to two physiological effects of food carbohydrates: postprandial glycaemia (post-meal elevation of blood glucose), and faecal bulking. Postprandial glycaemia is determined largely by carbohydrate digestibility,45 and faecal bulk largely by non-digestible, non￾fermentable polysaccharides.46 7.5.1 Limitations of carbohydrate composition data Standard food analyses do not account for the large effects of the structure of car￾bohydrate molecules and foods in the carbohydrate nutrition. Monosaccharide composition and order, glycosidic bonds, degree of polymerisation, chain con- figurations, non-covalent interactions between chains, and crosslinks that carbo￾hydrates readily form may all greatly affect physicochemical properties,6,47 and the physiological effects that depend on such properties. Furthermore, food struc￾ture, such as particle size, may considerably modulate the ability of food carbo￾hydrates to express their potential properties,35 by limiting solubility, extraction, and access of digestive enzymes. Beyond effects on extraction, interactions between carbohydrates and other food components in the intestine are multiple and complex.18 The amounts of carbohydrate fractions in foods are therefore not usually reliable guides to their physiological effectiveness.33,48 New approaches to providing nutritional information 173